Znorg. Chem. 1988, 27, 2733-2742
2733
Contribution from the Department of Chemistry, City University of New York, Queens College, Flushing, New York 11367
Photochemistry of Group 6B Hexacarbonyls Adsorbed onto Porous Vycor Glass. Evidence for the Formation of Methane from Carbon Species in the Glass Robert C. Simon, Edgar A. Mendoza, and Harry D. Gafney* Received October 8, 1987 Cr(C0)6, MO(C0)6, and W(CO)6 physisorb onto porous Vycor glass without disruption of the primary coordination spheres. UV photolysis of the adsorbed complexes leads to decarbonylation and formation of the corresponding pentacarbonyl, where the vacated coordination site is most likely occupied by a silanol oxygen. With 350-nm excitation, the quantum yields for decarbonylation of Mo(CO), and W(CQ6 are similar to those found in fluid solution, whereas that for Cr(C0)6 is significantly reduced. Continued photolysis with 350-nm light causes further decarbonylation,whereas 310- and 254-nm photolysis leads to CO, CH4, Hz, and COz evolution. CHI evolution initiates when the complex achieves an average molecularity of M(CO)4 and initially occurs in a 1:1 stoichiometric ratio with a concurrent oxidation of the mononuclear complex. The metal oxide agglomerates on the glass surface, and excitation of the agglomerate continues CHI evolution. Diffuse-reflectance FTIR spectra reveal formaldehyde and methanol intermediates. However, GC analysis and "C-labeling experiments establish that CHI does not derive from the initially coordinated CO. Rather, CH4 evolution is attributed to a metal-promoted hydrogenation of a C1 oxide within the glass matrix.
Introduction
Interest in the photoactivation of transition-metal carbonyls stems in part from their potential use as photocatalysts.' Wrighton and others, for example, have demonstrated the catalytic activity bf products derived from the photolyses of M(CO)6 ( M = Cr, Mo, and W) Although the majority of work has focus4 on photocatalytic behavior in homogeneous solution, recent studies have begun to explore the photoactivation of surface-confined metal carbonyl^^^^ as an alternate route to hybrid ~atalysts.'~Implicit in this approach, however, is an understanding of the photochemical behavior of a surface-confined metal carbonyl. Unless adsorption onto the support changes the complex, the primary photochemical event of a physisorbed metal carbonyl is equivalent to that in fluid On the other hand, the products derived from the reaction can be quite different from those obtained in fluid solution. Jackson and Thrushiem, for example, find that photolysis of Fe(CO)5 physisorbed onto silica gel yields Fe3(C0)1210whereas photolysis in the gas or liquid phases and in noncoordinating solvents yields Fe2(C0)9.'s-18 In contrast to the substitution and fragmentation reactions in homogeneous photolysis of R u ~ ( C O ) physisorbed '~ onto
(1) Wrighton, M. S. Ann. N.Y. Acad. Sci. 1980, 333, 188-207. (2) Wrighton, M. S.; Ginley, D. S.; Schroeder, M. A.; Morse, D. L. Pure Appl. Chem. 1975, 41, 671-679. (3) Wrighton, M. S.; Schroeder, M. A. J . Am. Chem. SOC.1973, 95, 5764-5765. (41 Wriahton, M. S.Inora. Chem. 1974, 13, 905-909. (5) Nashki, J.; Kirsch, k.;Wilputte-Steinert, L. J . Organomet. Chem. 1971, 27, C13-14. (6) Kinney, J. B.; Staley, R. H.; Reichel, C. L.; Wrighton, M. S. J . Am. Chem. SOC.1981, 103, 4273-4275. (7) Reichel, C. L.; Wrighton, M. S . J . Am. Chem. SOC.1981, 103,71807189; Inorg. Chem. 1980, 19, 3858-3863. (8) Liu, D. K.; Wrighton, M. S. J . Am. Chem. SOC.1982,104, 898-902. (9) Klein, B.; Kazlauskas, R. J.; Wrighton, M. S. Organometallics 1982, 1, 1338-1350. (10) Jackson, R. L.; Thrusheim, M. R. J . Am. Chem. SOC.1982, 104, 6590-6596. (11) Simon, R. C.; Gafney, H. D.; Morse, R. L. Inorg. Chem. 1983, 22, 573-574.
(12) Darsillo, M. S.; Gafney, H. D.; Paquette, M. S. J. Am. Chem. SOC. 1987, 109, 3275-3286.
(13) (14) (15) (16)
Dieter, T.; Gafney, H. D. Inorg. Chem. 1988, 27, 1730-1736. Bailey, D. C.; Langer, S. H. Chem. Rev. 1981, 81, 109-148. Graham. W. A. G.: Jetz. W. Inorp. Chem. 1971. 10. 4-9. Chisholm, M. H.; Masse;, A. G.; Thompson, N. R. iarure (London) 1966, 211, 67. (17) Dewar, J.; Jones, H. 0. Proc. R . SOC.London, Ser. A 1905, 76, 558-577. (18) Braye, E. H.; Huebel, W. Inorg. Synth. 1966, 8, 178-181. (19) Jackson, R. L.; Thrusheim, M. R. J . Phys. Chem. 1983.87, 1910-1916.
0020- 166918811327-2733$01.5010
porous Vycor glass leads to quantitative formation of the oxidative-addition product (~-H)RU~(CO)~~(~-OS~).~~ These changes in reactivity arise in the secondary thermal and/or photochemical reactions. Unlike a noncoordinating solvent, a hydroxylated
support, in particular, can function as a polydentate ligand, where binding to a surface functionality can stabilize a photoproduct, influence its spectral properties, and curtail its m ~ b i l i t y . ~ ~ JIn' J ~ addition, although less well understood at present, the dimensionality, Le., the irregularity of the surface, may be a further constraint on adsorbate reactivity.2' Each or any combination Qf these parameters will influence the reactivity of a surfaceconfined complex. To circumvent the difficulties imposed by the opacity of more traditional supports,'O we have examined the photochemical behavior of the group 6B hexacarbonyls on porous Vycor glass (PVG) and, in this paper, describe the results of these experiments. The similaritiesbetween PVG, a transparent, surfacehydroxylated silica with a myriad of randomly dispersed 70 f 21 A diameter pores, and silica gel have been d e s ~ r i b e d . ~ ~Irradiation -~~ of M(CO),(ads) (ads denotes the adsorbed complex) with 350-nm light causes a progressive decarbonylation of the complex, whereas irradiation with 310- or 254-nm light leads to the concurrent evolution of CH4, H2, and trace amounts of C02. However, GC analyses and %labeling experiments establish that CH4 does not derive from the coordinated CO. Rather, photoactivation of the complex promotes the hydrogenation of a C1 oxide within the glass matrix. Experimental Section Materials. Cr(C0)6, Mo(CO)~,and W(CO)6 (Pressure Chemical Co.) were used without further purification since their UV-visible and infrared spectra were in excellent agreement with published ~ p e c t r a . ~ ~ , ~ ~ W(13CO)6(Pressure Chemical Co.) was also used without further purification since its UV-visible spectrum also agreed with published spectra.25 The claimed enrichment (90%) was confirmed by mass spectroscopy and the integrated relative intensities (8.8:l) of the 1844- and 1986-cm-' "CO and CO bands. [(CzH5)4N][HW(CO),] was prepared under nitrogen according to the method of Gibson and co-workers?' and after several washings with hexane, the IR spectrum of the yellow-orange powder agreed with the reported ~pectrum.~'W 0 3 (Pfaltz and Bauer, 99.75%) and D20(Norell, 99%) were used as received. Gaseous reagents (Linde) were used without further purification since each had a purity (20) Desrosiers, M. F.; Wink, D. A.; Trautman, R.; Friedman, A. E.; Ford, P. C. J . Am. Chem. SOC.1986, 108, 1917-1927. (21) Pfeifer, P.; Avnir, D. J . Chem. Phys. 1983, 79, 3558-3565. (22) Iler, R. K. The Chemistry of Silica; Wiley-Interscience: New York, 1979; p 551. (23) Hair, M. L.; Chapman, I. D. J . Am. Ceram. SOC.1966, 49, 651-660; Trans. Faraday SOC.1965, 61, 1507-1510. (24) Cant, N. W.; Little, L. H. Can. J . Chem. 1964,42,802-809; 1965,43, 1252-1 254. (25) Wrighton, M. Chem. Reu. 1974, 74, 401-430. (26) Magee, T. A,; Matthews, C. N.; Wang, T. S.; Wotiz, J. H. J . Am. Chem. SOC.1961,83, 3200-3203. (27) Gibson, D. H.; Ahmed, F. U. Organometallics 1982, 1, 679-681.
0 1988 American Chemical Society
2734 Inorganic Chemistry, Vol. 27, No. 15, 1988
Simon et al.
Physical Measurements. UV-visible spectra were recorded on a Cary level of 299%. All solvents were HPLC grade (Fisher Scientific) and 14 or a Tectron 635 spectrophotometer adapted to accommodate the were dried over molecular sieves prior to use. rectangular cells. Adsorbate emission spectra were recorded on a preCode 7930 porous Vycor glass containing 70 f 21 A cavities was viously described Perkin-Elmer Hitachi MPF-2A emission spectrophoobtained from the Corning Glass Co. and used either as a fine powder tometer.2829J1Infrared spectra of solutions or KBr pellets were recorded (320 or 60-80 mesh) or as 25 mm X 25 mm X 4 mm plates. The glass on a Perkin-Elmer Model 1330 spectrometer calibrated with polystyrene. samples were continuously extracted, first withacetone for 12 h and then DRIFT spectra of crushed (320 mesh) impregnated PVG samples, diwith distilled water for an equivalent time. The extracted samples were luted 30:l with KBr, were recorded on a Nicolet 20/5DX FTIR instrudried under reduced pressure and then calcined either in air or under ment equipped with an intensified IR source, an MCTB detector mainflowing O2 or H2 at 650 'C for 272 h. The calcined samples were tained at 77 K, and a Harricks diffuse-reflectance accessory. The intransferred while hot, ca. 500 OC, either directly to a photolysis cell or terferograms were terminated under Happ-Ganzel apodization and to a vacuum desiccator and cooled to room temperature under vacuum converted to Kubelka-Munk spectra for quantitative comparison. All (p 5 lo4 Torr). spectra were ratioed against a background spectrum of blank, calcined Deuteriation was accomplished by refluxing either powdered or plate PVG diluted 30:l with KBr. Room- and low-temperature (77 K) EPR samples of the calcined glass in D 2 0 (Norell, 99%). After they were spectra were recorded on a IBM-Bruker 2ODE-SCR spectrometer. Mass refluxed for 8-12 h, the samples were dried under vacuum (p 5 lo4 spectra were recorded on a Varian MAT CH7 spectrometer calibrated Torr) at room temperature. Integration of the intensities of the 3744with perfluorokerosene or, in the low-molecular-weight region, with and 2760-cm-' bands of the silanol group and its deuteriated analogue known samples of CH, or C02. Solid samples of W('3CO)6 were subindicated that 33-50% of the Si-OH groups had been converted to Silimed into the spectrometer. Gaseous photoproducts were either bled OD. directly from the photolysis cell into the spectrometer or collected in a Impregnation. Impregnation was by conventional procedures in which sample loop that was then vented into the spectrometer. Both procedures calcined samples were exposed to 50 mL of a n-hexane solution 104-10-3 gave identical results. M in the hexacarbonyl. The moles of complex adsorbed, which ranged from lo-' to lo4 mol/g of PVG for the 25 mm X 25 mm X 4 mm pieces Results and as high as lo-) mol/g of PVG for the glass powder, was calculated Porous Vycor Glass. DRIFT spectra of PVG recorded as a from the change in optical density of the solution p h a ~ e .Hexane ~ ~ ~ ~ ~ function of temperature show that the three weight losses during incorporated during impregnation was removed under vacuum (p Ilo4 Torr) at room temperature. Removal of the solvent is a crucial point, thermal gravimetric analysis29 correspond to the loss of bulk water and the drying procedure was examined by diffuse-reflectance FTIR (100-120 "C) and chemisorbed water (2120 "C) up to the onset (DRIFT) and mass spectral measurements. DRIFT spectra recorded as of consolidation (ca. 700 "C). A t the calcination temperature, a function of time showed that the C-H bands of n-hexane disappeared 650 "C, spectral3 exhibit a sharp, intense band at 3744 cm-' with from the powdered glass after ca. 20 min of pumping (p Ilo-' Torr) a shoulder at 3655 cm-' that have been assigned to free and at room temperature. In the mass spectral measurements, a cell conhydrogen-bonded silanol groups, respecti~ely?~." Spectra of many taining a freshly impregnated 25 mm X 25 mm X 4 mm sample was of the calcined samples also show a weak, broad band a t 3500 attached to a vacuum line and pumped on. After various pumping times, cm-' that indicates the presence of chemisorbed water. The the cell was closed off, and after equilibration, the mass spectrum of the intensity of this band varies from sample to sample, but in all cases, vapor phase was recorded. No peaks indicative of n-hexane fragmentation or fragmentation of other hydrocarbons were detected for samples its intensity after calcination was 110%of that of the 3744-cm-' that had been pumped on for 230 min. Consequently, all samples imsilanol band. Deuteriated P V G exhibits an additional band a t pregnated by these solution procedures were pumped on for 230 min, and 2760 cm-' with a shoulder at 2641 cm-I that are assigned to the in some cases as long as 8 h, to ensure solvent removal. deuteriated analogues of the free and associated silanols, re: In the absence of a solvent, impregnation was by sublimation.12 A ~pective1y.I~A weak shoulder at higher wavenumbers, which @ weighed, calcined PVG sample was mounted upright in a rectangular attributed to the asymmetric stretch of DzO, again indicates the Pyrex cell and attached to a vacuum line. A 25-mL flask containing the presence of small amounts of chemisorbed D20in these samples. solid complex was attached to the line and the entire system evacuated. Adsorption of Cr(C0)6, Mo(CO)~,and w(Co)@Electronic A water bath (70 "C) was placed around the flask, and the compound spectra of the adsorbed complexes recorded ai different locations was allowed to equilibrate with the adjoining PVG sample. After it was equilibrated for various times, the flask was closed off and the cell and on a 25 mm X 25 m m X 4 mm sample are identical within line were reevacuated. The cell was then closed off, and the moles experimental error. The spectral agreement establishes that imadsorbed onto the PVG was determined by spectral comparison with pregnation by either solution adsorption or sublimation leads to samples impregnated by the solution procedures. a uniform distribution of the hexacarbonyls on the PVG surPhotochemical Procedures. Impregnated pieces of PVG were rigidly face.28,29,31Regardless of the moles adsorbed, however, each mounted with a Teflon holder in previously described rectangular quartz carbonyl complex penetrates no more than 0.5 f 0.1 m m and or Pyrex ~ e l l s . ~The J ~cell and enclosed sample were evacuated on a line impregnates regions adjacent to the outer geometric surfaces of equipped with an oil diffusion pump isolated from the line by a trap at the glass.11,28,29 A linear decline in adsorbate optical density with 77 K. Unless otherwise specified, all samples were irradiated under thickness, i.e., optical path length, within the impregnated volumes vacuum (p 5 lo4 Torr) in a Rayonet reactor (Southern New England Ultraviolet Corp.) equipped with 254-, 310-, or 350-nm bulbs. The implies a uniform distribution of the complexes within these photon flux, typically (2-4) X lo9 einstein/(cm2 s), was measured with volumes.28 Taking the area occupied by M(C0)6(ads) ( M = Cr, a ferrioxalate solution in a rectangular cell of the same dimension^.^^ Mo, W ) to be 38.5 A232 and the surface area to be 130 m2/g,29 Quantum yield measurements were limited to 350-nm excitation since the samples prepared in these experiments, which ranged from competitive absorption by the glass (50% transmittance (T) at 295 nm lo-' to mol/g of PVG, correspond to fractional surface vs air) is negligible at this wavelength. UV-visible spectra were recorded coverages within the impregnated regions of 1O.OOO8 to essentially during photolysis, and the initial rates of product formation or reactant monolayer coverage. Since the complex is not uniformly disconsumption were calculated according to previously described procetributed throughout the entire sample, however, the calculated d u r e ~ Scattering .~~ from the glass surface as well as uncertainties in the values are approximations and a r e used in a relative rather than optical pathlength within the glass, however, introduce an estimated uncertainty of 25% in reported quantum yields. absolute sense. Gaseous photoproducts were collected and analyzed by previously Spectral Properties of Cr ( c o ) 6 (ads), Mo( Co),( ads), and described GC pro~edures.'~J~ The GC analyses were calibrated by the W(CO),(ads). The electronic spectra of the adsorbed complexes addition of known pressures of each gas to the photolysis cell. Collection In closely resemble spectra of the complexes in t~-hexane.l'-~'+~~ and transfer of the gas was accomplished with a Toepler pump, and the the 500-260-nm region, spectra of Cr(CO)6(ads), M ~ ( C o ) ~ ( a d s ) , detector response was linear in the moles of gas in the cell. In the isotopic and W(C0)6(ads) show intense bands centered at 280,286, and labeling experiments, the gaseous photpproducts were either collected in 287 nm, respectively. The lower energy, ligand field transitions, the same manner or transferred directly to a mass spectrometer. (28) Wolfgang, S.; Gafney, H. D. J . Phys. Chem. 1983, 87, 5395-5401. (29) Kennelly, T.; Gafney, H. D.; Braun, M.J. Am. Chem. Soc. 1985,107, 443 1-4440. (30) Calvert, J. G.;Pitts, J. N. Photochemistry; Wiley: New York, 1971; pp 780-788.
(31) Basu, A,; Gafney, H. D.; Perettie, D. J.; Clark, J. B. J . Phys. Chem. 1983,87,4532-4538. (32) Beach, N. A.; Gray, H. B. J . Am. Chem. SOC.1968, 90, 5713-5721. (33) Geoffroy, G. L.; Wrighton, M. S. Organometallic Photochemistry; Academic: New York, 1979; pp 46-49.
Group 6B Hexacarbonyls Adsorbed onto Vycor Glass
Inorganic Chemistry, Vol. 27, No. 15, 1988 2735 Table I. Quantum Yields for Decarbonylation during 350-nm Photolysis and Estimates of the Quantum Yields of CH4 Evolution during 3 10- and 254-nm Photolyses
hib complex baa 310nm 254nm 0.081 0.025 0.03 0.002 Cr(CO), Mo(CO), 0.79 & 0.10 0.10 0.008 0.84 0.09 0.14 0.01 W(CO), "Quantum yield for reaction 1. bEstimated uncertainty of 125% (see Experimental Section).
* *
reaction stoichiometry and the similarity of the photoproduct spectra with those of the corresponding pentacarbonyls generated in low-temperature mat rice^'^.^^ establish the primary photochemical event to be M(CO),(ads)
300 I
400
I
I
nm
Figure 1. Spectral changcs during 350-nm photolysis of 2.57 X mol of W(CO),(ads)/g of PVG in vacuo. Numbers designate photolysis timcs in seconds. which appear as low-intensity shoulders on the above chargetransfer transitions in solution spectra,33appear as an unresolved tail in the 300-350-nm region. The molar extinction coefficients of the adsorbed complexes, calculated by assuming an optical path length equivalent to the penetration depth, are 15-20% larger than those in n-hexane solution. The actual difference is most likely smaller since there is an inherent spectroscopic uncertainty of 58% from sample to sample and a 20% uncertainty in penetration depth.29 The latter describes cross-sectional distribution in the general sense but is not an accurate measure of the optical path length. DRIFT spectra of Cr(CO),(ads), M ~ ( c O ) ~ ( a d sand ), W(CO),(ads) show intense bands centered at 1999, 2005, and 1986 cm-', respectively. With the exception of some broadening, particularly the lower energy shoulders, these spectra also resemble the spectra of the complexes in n-hexane." Photochemical Reactiom of Cr(CO),(ads), Mo(CO)6(ads), and W(cO),(ads). Excitation with 350-nm light causes an immediate decline in the UV absorption characteristic of each hexacarbonyl and a concurrent growth of bands in the 240-250- and 400450-MI regions. Figure 1, which is representative of each complex, illustrates the spectral changes during a 350-nm photolysis of 2.57 X lo-' mol of W(CO),(ads)/g of PVG in vacuo. Maintainance of the isosbestic points and the reversibility of the reaction, however, vary from complex to complex. The bright yellow of Cr(CO),(ads), due to an absorption at 448 nm, fades considerably in the time required to transfer the cell to the spectrophotometer, whereas the intensity of the 410-run absorption of W(CO)5declines 14%, after photolysis. With W(CO),(ads), the isosbestic point at 358 nm (Figure 1) is maintained through 40% reaction. With Cr(CO),(ads), the isosbestic point occurs at 360 nm and is maintained through 115% reaction, while that for Mo(C0)6(ads) exists only for the first few percent of reaction. With each complex, the initial spectral changes are independent of surface coverage, and if the isosbestic point is maintained, G C analysis of the cell contents after 350-nm photolysis in vacuo yields 1.0 f 0.1 mol of CO/mol of M(CO), reacted; the latter is determined from the decrease in UV absorbance of the hexacarbonyl. The
-
M(CO),(ads)
+ CO
(1)
Perutz and Turner attribute the changes in band maxima of the individual pentacarbonyls generated in low-temperature matrices to an occupation of the vacated coordination site by the different matrix atoms or molecules.34 The band maxima of the photoproducts on PVG resemble those of M(CO),L complexes, where L denotes an 0-donor ligand. For example, the band maximum of W(CO)5(ads) occurs a t 410 nm, while those of W(CO),(Me2CO) and W(CO)5(Et2CO) occur a t 406 and 418 nm, respectively.2s Since the surface functionalities of a hydroxylated support are potential nucleophiles, M(CO)5(ads) denotes a pentacarbonyl with the vacated coordination site occupied by an oxygen atom from either a silanol group or chemisorbed water. The absence of EPR resonances after 15-35% conversion to the photoproducts suggests that M(CO),(ads) has a C, structure." If the isosbestic point is maintained during photolysis, reaction 1 is reversible, and exposure to CO (1 atm) regenerates the hexacarbonyl in 196% yield. In the presence of O2 (1 atm), photolysis of the adsorbed complexes leads to irreversible oxidation. A rapid evolution of 5.9 f 0.2 mol of CO/mol of adsorbed complex occurs during photolysis, and there is no spectrally detectable reversibility on exposure to CO (1 atm). The isosbestic points observed during 350-nm photolyses are not present during 3 10- and 254-nm photolyses. However, the initial spectral changes, which consist of the immediate growth of bands in the 240-250- and 400-450-nm regions, indicate that the primary photochemical reaction of each adsorbed complex, Le., reaction 1, is independent of excitation wavelength. Accurate quantum yield measurements are restricted to 350-nm excitation, where absorption by PVG is negligible. The values for reaction 1, extrapolated from plots of &,M (corrected for thermal reversibility by extrapolating the spectral change back to the end of photolysis) vs irradiation time, are summarized in Table I. The primary photochemical reaction is independent of the excitation wavelength, whereas the secondary photochemistry exhibits a marked dependence on excitation wavelength. Although the following focuses on W(CO),(ads), since most experiments were performed with this complex, the reactivities of the Mo and Cr analogues are essentially equivalent. Continued 350-nm photolysis of M(CO),(ads) causes further C O evolution. With W(CO),(ads), the yield of co approaches the limiting value 1.1 i 0.2 mol of CO/mol of hexacarbonyl adsorbed. The stoichiometry suggests a straightforward formation of the pentacarbonyl; however, electronic spectra indicate a more complex reaction sequence. A loss of the isosbestic point at 358 nm is accompanied by a decline and nondescript broadening of the photoproduct absorption at 410 nm. With samples containing ?lo" mol of M(CO),(ads), there is no spectral indication in the 475-550-nm region of the formation of dimeric or higher order cluster corn pound^.^^-^^ Most likely, the stoichiometric mea(34) Perutz, R. N.; Turner, J. J. J. Am. Chem. Soc. 1975,97, 4791-4800. (35) Graham, M.A.; Poliakoff, M.;Turner, J. J. J . Chem. Soc. A 1971, 2939-2948.
2136 Inorganic Chemistry, Vol. 27, No. 15, 1988
- 8 -
8
10
I
scc x 103
Figure 2. Yields of CO ( O ) , CH4 (A),H2 (a),and W(CO)s(ads) (B) during 310-nm photolysis of 8.50 X 10“ mol of w(co)6/g of PVG.
15
/
5
10
sec x 10-5
Figure 3. Yields of CO (A), CHI ( O ) , and H2 (4) during 254-nm photolysis of 2.35 X mol of w(Co),/g of PVG.
surements are fortuitous, and the spectral broadening, although not resolved, corresponds to a composite spectrum of different monomeric M(CO), (n 1 5 ) species.34 Nevertheless, the photochemistry with 350-nm excitation is limited to C O evolution. Methane Evolution. Photolysis with 310- or 254-nm light leads to different reaction products. A 310-nm photolysis of a sample containing 8.50 X l o d mol of W(cO),(ads)/g of PVG in vacuo (Figure 2) causes CO, Ha, and CH4 evolution. Similar results occur with 254-nm excitation (Figure 3) except that the initial spectral changes indicative of M(CO)5(ads) formation are not as well resolved. With both excitation wavelengths, C 0 2 is also detected, but in trace amounts corresponding to 110% of the amount of CHI evolved. As described above, CH, evolution cannot be attributed to a decomposition of trace amounts of solvent, n-hexane, remaining in the glass. Furthermore, photolysis of samples prepared by solventless techniques, i.e., by sublimation of the complexes onto freshly calcined PVG samples, leads to CH4 evolution within experimental error of that shown in Figure 2. Nor can CH, evolution be attributed to direct excitation of PVG (50% T at 295 nm vs air). PVG absorbs weakly at 310 nm and, particularly with less than monolayer coverage, is the dominant (36) Wrighton, M. S.;Ginley, D. S. J . Am. Chem. Soc. 1975, 97, 4246. (37) Ginley, D. S.;Bock, C. R.; Wrighton, M. S. Inorg. Chim. Acra 1977, 23, 85.
Simon et al. absorbing species at 254 nm. However, 3 10- or 254-nm photolysis of blank PVG under vacuum, under 0.1 atm of CO or H2,or under a mixture of 0.1 atm of CO and 0.2 atm of H2 does not result in CH4 evolution. Competitive absorption of the 310- and 254-nm excitation by the glass precludes an exact determination of the quantum yield of CH4 evolution. To give some indication of the relative extent of this reaction pathway, however, estimates of the quantum yields of CH4 evolution, +M, computed from the initial rates of CH, evolution and the excitation intensity incident on the impregnated sample are listed in Table I. The lower values with 254-nm excitation, where a larger fraction of the excitation light is absorbed by the glass, are consistent with the observation that direct excitation of the unimpregnated glass does not lead to CH4 evolution. The complex is essential to CH4evolution, and the CH, does not derive from n-hexane. The absence of CH4 evolution during the initial photochemical reaction (Figure 2) precludes the primary photoproduct species as the active reagent. During 254-nm photolysis, the isosbestic points are not present and the spectral changes indicative of M(CO)5(ads) formation are not as well resolved. However, extrapolations of the initial rates of methane evolution (Figure 3) reveal a short induction period preceding H2 and CH, evolution. CO evolution continues during the initial stages of the reaction, but eventually CH4 becomes the dominant gaseous photoproduct. The rates of CH,, H2, and COz evolution vary even within samples containing the same amount of complex, but the order of appearance of the gaseous photoproducts (Figures 2 and 3) is consistent from sample to sample and independent of the initial amount of adsorbed hexacarbonyl. A 310- or 254-nm photolysis of the complexes adsorbed onto deuteriated PVG leads to the evolution of deuteriated methanes. Mass spectral analyses of the surrounding gas phase after 3 IO-nm photolyses of deuteriated PVG samples containing (5.2 f 0.1) X 10” mol of W(CO)6 reveal CH4, CH3D, and small amounts of CH2D2. The amount of deuteriated products increases with the extent of deuteriation of the glass (3340%) but never exceeds 25% of the amount of CHI evolved. Thermal gravimetric analyses show that water is tenaciously held on PVG:9 and a weak, high-frequency shoulder on the 276km-l band of Si-OD indicates the presence of small amounts of chemisorbed DzO. Since a rapid exchange between the various hydroxyl moieties is expected, these data only designate the silanol group and/or chemisorbed water as the hydrogen source. CH, evolution is independent of surface coverage and occurs from samples containing from IO-’ to mol of W(CO)6(ads), which correspond to surface coverages ranging from 0.0008 to essentially monolayer coverage within the impregnated volumes. Nor is there spectral evidence of dimer formation or an intermediate metal hydride. A 3 10-nm photolysis of a sample containing 9.26 X 10” mol of W(C0)6(ads) causes the evolution of 7.54 X 10” mol of CH4 Assuming that the spectrum of a dimeric species on the glass resembles that of [W2($-C5H5)2(CO),], which exhibits bands at 362 nm (t = 2.02 X lo4 M-’ cm-’ ) and 493 nm ( e = 2.45 X lo3 M-’ cm-’ ),36 the spectral changes at 358 nm, the initial isosbestic point (Figure l), and at 493 nm indicate that 5 10-8 mol of dimer is formed. Similarly, there is no spectroscopic indication of the formation of dimeric complexes during photolysis of the Cr and Mo analogue^.^' In fluid solution, [(CzH5)4N][HW(CO),] exhibits a band at 450 nm (e = 4.6 X lo4 M-’ cm-’ ) and a similar band when adsorbed onto PVG. During CH4 evolution, the 410-nm band of W(CO)5 declines and shows a nondescript broadening. A distinct band at 450 nm does not appear, and the optical density changes at 450 nm indicate that 510-9 mol of hydride is present. Furthermore, 254-nm photolysis of a PVG sample containing 1.3 X 10” mol of [(C2H5),N][HW(CO),]/g of PVG in vacuo does not result in either a spectral change or the evolution of gaseous photoproducts. Initially, H2 evolution accompanies CH4 evolution (Figures 2 and 3). H2evolution suggests metal oxidation, and EPR confirms its occurrence. EPR spectra recorded during 254-nm photolysis of crushed (60-80 mesh) PVG containing 7.83 X 10” mol of
Inorganic Chemistry, Vol. 27, No. 15, 1988 2737
Group 6B Hexacarbonyls Adsorbed onto Vycor Glass
2 I
I
I
3200
3500
I
I
3800
4100
I
10 I
I
mdesCH4x106
Figure 5. Moles of W(CO)6 irreversibly oxidized, n, - n, (see text), vs the moles of CH, evolved.
Gauss
Figure 4. EPR spectrum generated during 254-nm photolysis of 7.83 lod mol of Mo(C0)6(ads)/gof crushed (60-80 mesh) PVG.
6 I
X
M o ( C O ) ~in vacuo results in the appearance (Figure 4) of a resonance with g = 1.948 f 0.002 and peak width of 75 G. No resonance occurs when blank, calcined PVG (60-80 mesh) is irradiated under identical conditions. The signal intensity increases with increasing irradiation time, and GC analysis of the surrounding gas phase confirms the presence of CH,. The EPR resonance is in excellent agreement with that described by Howe and Leith and establishes the formation of the Mo(V) hydrous oxide, M o O ~ ( O H ) . EPR ~ ~ resonances are not detected during similar experiments with Cr(C0)6(ads) and W(C0)6(ads). After extensive 3 10- or 254-nm photolysis of the latter, however, there is no reversibility when it is exposed to CO and electronic spectra are consistent with metal oxidation. The W(C0)6(ads) samples become light blue and the electronic spectra, having a sharp spectral cutoff with 50% T at 465 nm, are essentially identical with spectra of W 0 3 powder dispersed on the glass. Stoichiometric Measurements. Since the oxidized metal will not react with CO, the stoichiometryof the oxidation accompanying CHI formation was determined by recovering the unoxidized metal as the hexacarbonyl. The number of moles of CH4 evolved was determined by gas chromatography while the moles of hexacarbonyl consumed during photolysis of a sample containing 4.02 X lod mol of W(CO),/g of PVG, n,, was determined from the decline in the UV absorbance of the complex. The sample was then exposed to CO (1 atm), and the moles of hexacarbonyl regenerated, 4,was calculated from the spectral change at 280 nm. A plot of the moles of complex irreversibly oxidized, n, n,, vs the moles of CH, evolved (Figure 5) shows that oxidation of the metal initially occurs in a 1:l stoichiometric ratio with the moles of CH4 evolved. At the onset of CHI evolution during 310-nm photolysis of a sample containing 9.77 X 10" mol of W(CO)6(ads),a total of 1.73 x mol of CO has been evolved. This corresponds to an average loss of 1.8 mol of CO/mol of complex reacted. A series of measurements with samples containing from 5 X l@ to 5 X mol of W(CO)6 yield an average adsorbate stoichiometrv at the onset of CHd evolution of W(co)4.0+0.2* Diffuse-Reflectance FTIR. DRIFT spectra recorded during 254-nm photolyses of PVG (325 mesh) containing 2.06 X mol of W(CO,k(ads) (Figure 6a) show that the growth of weak bands at 1965 and. 1932 cm-' accompanies the decline in the intensity of the 1988-cm-' band of hexacarbonyl. The 1965- and 1932-cm-' bands agree with the intense 1967- and 1926-cm-' bands of W(CO)s generated in a 20 K CHI matrix3, and are (38) Howe, R. F.;Leith, I. R. J . Chem. Sa.,Furuday Trans. 1973, 69, 1967-1 977.
assigned to surface-bound pentacarbonyl. Consistent with electronic spectra, which indicate efficient secondary photolysis, the intensities of the 1965- and 1932-cm-I bands of the pentacarbonyl increase only during the first 15 s of photolysis. Continued photolysis causes these bands to decline with the accompanying appearance of weak bands at 1949 and 1894 cm-'. Although weak, these features appear in each experiment and are in reasonable agreement with the prominent 1945- and 1887-cm-' bands of W(CO)4.28 The bands assigned to the tetracarbonyl do not increase on continued photolysis. Instead, the entire urn region degrades without the appearance of lower wavenumber bands assignable to W(CO), (n I 3) species.34 The decline in band intensity in the CO region, which eventually becomes transparent, is accompanied by the appearance of bands (Figure 6b) in the initially transparent C-H region. After 15 s, difference spectra reveal an absorption centered at 2964 cm-'.The band intensity increases during photolysis and, after 45 s, resolves into peaks at 2967 and 2961 cm-'. A distinct shoulder is also apparent at 2940 cm-' as well as a broad absorption in the 3210-3450-crn-' region. Other than changes in band shape, Le., the 2967- and 2961-cm-' bands often appeared as a broader absorption centered at 2965 cm-', no further change in band intensity occurs during an additional 30-min photolysis. The strong absorptions due to PVG preclude detection of bands at I 1 900 cm-'. In the high-wavelength region, however, the photoproduct spectrum agrees with the spectrum of formaldehyde adsorbed onto PVG, which exhibits a strong absorption at 2966 cm-' with a shoulder at 2945 cm-I and a broad absorption in the 32103450-cm-I region. Isotope Labeling. Since the surface of PVG closely resembles that of silica ge122-24and the gaseous photoproducts in these experiments as well as their order of appearance parallel those found in the temperature-programmed decomposition of these hexacarbonyls on silica gel,39340we initially assumed that CH4 derived from coordinated CO with an accompanying oxidation of the metal atom. However, CH, evolution becomes independent of metal oxidation (Figure 5). In fact, GC analysis after exhaustive 254-nm photolysis of a sample containing 2.30 X mol of W(CO)6 reveals, in addition to 2.52 X lo4 mol of CH,, 1.35 X lo4 mol of CO, which corresponds to a recovery of 98% of the CO initially coordinated to the metal. A series of experiments with samples containing from 6.31 X lo4 to 4.33 X 10" mol of W(CO),(ads)/g of PVG confirmed that the amount of CO recovered corresponded to 94-9896 of the CO originally coordinated to the metal. This raises a question regarding the source (39) Brenner, A.; Hucul, D. A.; Hardwick, S. J. Inorg. Chem. 1979, 18, 1478-1 484. (40) Brenner, A.; Hucul, D. A. J. Am. Chem. Soc. 1980, 102, 2484-2487.
2738 Inorganic Chemistry, Vol. 27, No. 15, 1988
Simon et al.
la
A
b
JUUJ
2076
2864
2000
1926
WAVEN!JMBEM (Cn-’)
Figure 6.
DRIFT spectra during 254-nm photolysis of 2.06 X lo-$ mol of W(CO),(ads)/g of powdered (320 mesh) PVG. Numbers in parentheses
indicate photolysis times in seconds.
of CH4, and 13Clabeling confirms that CH4 does not derive from the coordinated CO. After 310-nm photolyses of PVG samples containing 3.62 X 10” mol of W(13CO)6(90% enriched), mass spectral analysis shows that 5 1 % of the 13Clabel appears in the evolved CHI. Rather, 96 f 1% of the label appears as evolved 13C0,which occurs in the expected 9:l ratio, with the remaining 4 f 1% appearing as evolved C02. Nineteen experiments with separate samples containing from 7.51 X l@ to 5.30 X mol of labeled W(13CO)6excited with either 254- or 310-nm light established that in all cases 51% of the label occurs in the evolved CH4. The result is independent of whether the samples were initially prepared by solution impregnation or sublimation, Le., solventless techniques. Clearly, the evolution of CHI derives from an impurity, yet it is not one that can be attributed to conventional impurity sources, such as residual amounts of the impregnating solvent, n-hexane, or adsorption of trace oils in the vacuum lines during sample manipulations. In the first place, CH4 evolution occurs from freshly calcined samples impregnated by solventless techniques. Furthermore, both impurity sources would be expected to show, in addition to gaseous photoproducts, either the fragmentation pattern characteristic of n-hexane or peaks in the higher m/e range arising from the decomposition of an adsorbed oil. Yet neither is detected. To further test the possibility of contamination, a freshly calcined sample at ca. 500 OC was transferred to a photolysis cell with a side arm containing W(CO)6. The portion of the cell containing the glass sample was placed in a sand bath at 250 OC, and the entire apparatus, including the solid complex at room temperature (22 f 1 “C), was evacuated t o p Il@ Torr. The cell was then closed off, and after the system was cooled to room temperature, the complex was sublimed onto the glass sample. Again, mass spectra of the gas phase after 254-nm photolysis showed the presence of only CO, H2, CH4, and COz. In all experiments, including those with the labeled complex, mass spectra of the surrounding gas show that unlabeled CH, is the sole hydrocarbon detected, and the highest mle peak in the spectra occurs at 45, which corresponds to 13C02. Since CH, evolution occurs from samples that had been calcined at 650 OC for at least 72 h, and in some cases as long as 1 week, CH4 does not derive from an adsorbed hydrocarbon. Rather, since the C source withstands the calcination conditions, we believe that CH,
evolution arises from a metal-promoted reduction of a CI oxide present in the glass Burnham reports that the principal form of C in silica melts is C032-, which derives from the reaction of C 0 2 with dangling 02-functions of the melt.43 Although analysis of the calcined PVG with an Hitachi SEM equipped with a PGT energy dispersive X-ray analysis accessory failed to resolve that attributable to the impurity from residual background carbon, elemental analysis of calcined PVG indicates trace amounts, SI%, of C that is thought to be a C1 oxide in the glass matrix.44 Prolonged photolysis over a period of days shows that the amount of CH4 evolved falls within this impurity level, and upon consumption of the C impurity further gas evolution ceases. For example, prolonged 254-nm photolysis of a PVG sample containing 2.30 X 10” mol of W(CO)6 in vacuo leads to the evolution of 1.35 X lo4 mol of CO, which corresponds to recovery of 98% of the originally coordinated CO. CH4 evolution increases during the initial stages of photolysis, achieves a steady-state rate, and then declines and ceases with no further gas evolution. The complex penetrates 0.5 f 0.1 mm into the glass, which corresponds to an impregnation of 20% of the PVG sample. Assuming that the C impurity converted to CH4 is that within these initially impregnated volumes, the total moles of CH4 evolved in the experiment, 2.52 X lo4 mol, corresponds to a C impurity level, 0.34 f 0.07%, that lies within the reported impurity levels.44 Discussion In fluid solution, UV photolysis of the group 6B hexacarbonyls results in CO dissociation and in the presence of a nucleophile, L, formation of M(C0)5L. Studies in low-temperature matrices confirm the formation of the corresponding p e n t a c a r b o n y l ~ ~ ~ , ~ ~ and support a reaction sequence where the primary photochemical step is CO dissociation followed by tapid scavenging of the pen(41) Reference 22, p 3. (42) (a) Dake, H. C.; Flcener, F. L.; Wilson, B. H. Quwtz Family Minerals; McGraw-Hill: New York, 1938. (b) Cameron, E. N.; Row, R. B.; Weis, P. L. Am. Mineral. 1953, 38, 218-262. (c) Roedder, E. Reo. Mineral. 1984, 12, 525. (43) Burnham, C. W. In The Euolution ofthe Igneous Rocks;Yoder, H. S., Ed.; Princeton University Press: Princeton, NJ; 1979; p 454. (44) Morse, D. L., private communication, Corning Glass Works, 1984.
Group 6B Hexacarbonyls Adsorbed onto Vycor Glass tacarbonyl by an available nucleophile. Strohmeier and co-workers report initial quantum yields for M(C0)5L formatian, ca. 1, that indicate a primary photochemical event of high e f f c i e n ~ y ? ~On the other hand, Nasielski and Cola report a limiting quantum yield for the photosubstitution of CO by pyridine in Cr(C0)6 of 0.67 f 0.2.& In spite of the uncertainty, it is agreed that a relatively efficient loss of CO from these complexes is followed by rapid nucleophilic addition to the pentacarbonyl?* DRIFT' spectra of calcined PVG indicate a surface composed of free and associated silanol groups as well as small amounts of chemisorbed water. Chemically, the surface resembles that of silica gel except that Lewis acid sites, principally B2O3, are also resent.^^-^' However, spectral data do not indicate any preferential adsorption of these or other complexes onto these sites.'213" Adsorption, either from fluid solution or by sublimation, is a random process leading to a uniform distribution of the hexacarbonyls on the surface. As found with other metal complexes,12*13,28.29J1 the hexacarbonyls penetrate 0.5 f 0.1 mm into the glass, and the penetration depth is taken as a measure of the deviation from surface planarity rather than population of the interior cavities of the glass. Consequently, the photochemical reactions examined in these experiments are limited to molecules on the outer, albeit highly i r r e g ~ l a r ?surfaces ~ of PVG. The absence of decomposition products, such as CO or H2, during adsorption as well as the similarity of spectra of the adsorbed complexes with fluid-solution spectra establish that the complexes physisorb onto PVG as distinct molecular entities without disruption or significant distortion of their primary coordimtion Therefore, changes in photoreactivity relative to that found in fluid solution cannot be attributed to a molecular change in the adsorbed complex. In fact, the initial photoreactions of the adsorbed complexes, particularly those during 350-nm excitation, exactly parallel that found in fluid solution?* The evolution of 1.0 f 0.1 mol of CO/mol of complex consumed, the similarity of the photoproduct spectra (Figure 1) with spectra of the corresponding pentacarbonyls generated in low-temperature mat rice^,^^.^^ and the recovery of 196% of the hexacarbonyl on exposure to CO (1 atm) esta'blish reaction 1 as the primary photochemical event. Although isosbestic points are not observed during 310- and 254-nm photolysis (see below), the similarity of the initial spectral changes with those during 350-nm photolysis indicates that the primary photoreaction, i.e., reaction 1, is independent of excitation wavelength. The absence of EPR resonances during 350-nm photolysis and the similarity of the photoproduct spectra with those of known M(CO),(O-donor) complexes are consistent with the singlet ground state of a square-pyramidal structure with the vacated coordination site occupied by an oxygen atom." In spite of the relatively small amount of chemisorbed water on the calcined glass, recent studies of physisorbed Fe(CO), indicate that the primary, tetracarbonyl, photoproduct reacts with either chemisorbed water or a silanol group, the latter being the more stable product.I2 Most likely, similar reaction pathways are accessible to the pentacarbonyls generated in these experiments. Both pentacarbonyl adducts would be expected to exhibit similar electronic spectra with bands in the 400-450-nm region. However, the short lifetime of the Fe(C0)4.H20 adduct, I 1 0 0 s in vacuo,12suggests that the photoproduct spectra recorded in these experiments (Figure 1) are principally those of the silanol adduct. Consequently, M(CO)5(ads)designates a C , species where the silanol group binds the pentacarbonyl to the glass. Whether this represents a formal coordination to the glass, however, is not clear. Perutz and Turner report that the electronic spectra of the pentacarbonyls generated in Ar, CH4, and N2 matrices are sensitive to the matrix atom or molecule occupying the vacated coordination site.M Since the latter must be regarded as very weak ligands, the similarity with spectra (45) Strohmeier, W.; Gerlach, K. Chem. Ber. 1961, 94, 398-406. (46) Nasielski, J.; Cola, A. J. Orgummet. Chem. 1975, 101, 215-219. (47) Evan, U.; Rademan, K.; Jortner, J.; Manor, M.; Reisfeld, R. Phys. Rev. Lcff. 1984, 52, 2164-2167. (48) Reference 33, p 69.
Inorganic Chemistry, Vol. 27, No. 15, 1988 2739 of W(C0) ,(O-donor) complexes does not necessarily imply, in our opinion, a formal coordination to the glass. Furthermore, the activation energy, 1 7 kcal/m01?~for the thermal back-reaction W(CO),(ads)
+ CO
-
W(CO)6
(2)
is considerably smaller than that for the substitution reactions of W(CO)5L complexes in fluid solution.49 If overcoming the W(CO),-PVG interaction is the rate-determining step, then the activation energy suggests an interaction that, at least energetically, is significantly less than formal coordination to the silanol group. Of course, any attempt to draw thermodynamicconclusions from kinetic data is suspect. Nevertheless, the interaction with the glass stabilizes the primary, pentacarbonyl, photoproduct,11and as found in fluid solution, where the quantum yield of decomposition reflects a competition between addition of CO and an available nucleo~ h i l e ?the ~ quantum yields of decomposition of the adsorbed complex (Table I) reflect a competition between addition of CO and the silanol group. Although dependent on previous heatings, studies of a variety of hydroxylated silicas indicate 4-7 silanol groups/ 100 A2.51*52 If 3.5 A is taken as the radius of M(CO)6,32the planar surface area covered by the adsorbed carbonyl, 38.5 A2,places one to three silanol groups in the immediate vicinity of the photoproduct. Therefore, the quantum yields of formation of M(CO),(ads) (Table I) are not limited by the availability of silanol groups. In fact, the yields for Mo(CO),(ads), 0.79 f 0.10, and W(CO),(ads), 0.84 f 0.09, are comparable to the reported yields in fluid solut i ~ n . However, ~ ~ , ~ that for Cr(CO),(ads), 0.081 f 0.02, is significantly smaller than the reported solution value, 0.67 f 0.2.46 This difference in yields is not due to a cage effect or a change in the adsorbate's photophysical properties. The electronic spectrum of Cr(C0)6(ads), like that of the Mo and W analogues, is equivalent to that in fluid solution." Since the cross-sectional distributions of each complex in the glass are identical within experimental error, a cage effect53that might arise from the low dimensionality of the PVG surface47would be independent of the metal complex, which is clearly not the case. Rather, the difference in yield must reflect a much weaker interaction between Cr(CO),(ads) and the silanol group. Unlike the case for Mo(CO), (ads) and W (CO) 5( ads), which show little reversibility, the significant decline in the 448-nm band of Cr(CO)5(ads) and concurrent growth of the 280-nm band of Cr(C0)6(ads) after photolysis indicate that the pentacarbonyl thermally reacts with the photodetached CO. A weaker interaction with the glass surface would facilitate a thermal reaction with the photodetached CO even after separation of the reaction pair and, in spite of the corrections for thermal 'reversibility, bias the subsequent calculation of the yield. Reaction 1 is independent of excitation wavelength, whereas the secondary photochemistry is wavelength-dependent. With 350-nm excitation, CO evolution occurs, and in spite of a limiting stoichiometry of 1.1 f 0.2 mol of CO/mol of W(C0)6(adS) reacted, a loss of the isosbestic point and a decline and nondescript broadening of the initial 'A 'E absorptions of the pentacarbonyl suggest the formation of M(CO),(ads) ( n I5 ) species.34 The spectral change is independent of surface coverage and occurs without the appearance of lower energy bands attributable to polymeric species. The point to be emphasized, however, is that 350-nm excitation causes only CO evolution, whereas continued photolysis with 310- or 254-nm light leads to CO, H2, CH4, and C 0 2 evolution (Figures 3 and 4). The induction periods in Figures 2 and 3 establish that the pentacarbonyls are photochemically converted to a species that mediates CHI evolution. Stoichiometricmeasurements indicate that CH4 evolution occurs when the complex achieves an average
-
(49) Simon, R. C.; Morse, D. L.; Gafney, H. D. Inorg. Chem. 1985, 24, 2565-2570. (50) Angelici, R. J. Organomet. Chem. Rev., Secr. A 1968, 3, 173-226. (51) Snyder, L. R.; Ward, J. W. J. Phys. Chem. 1966, 70, 3941-3952. (52) Reference 22, pp 622-714. (53) Massey, A. G.; Orgel, L. E. Nature (London) 1961, 191, 1387.
2740 Inorganic Chemistry, Vol. 27, No. 15, 1988 molecularity of W(CO),. The molecularity is clearly an average, but it is consistent with the temperature-programmed decomposition of these complexes on silica gel39poand the absence of lower frequency bands attributable to W(CO), (n I3) species" during CH4 evolution. In the temperature-programmed decomposition of these complexes on silica gel, Brenner and co-workers report that 2 mol of CO/mol of complex is evolved prior to the evolution of CH4.39940Since DRIFT spectra of a number of Fe(CO), (n I4) species on PVG have been recorded,12we believe that the absence of lower frequency bands attributable to W(CO), (n 5 3)', is not due simply to their low steady-state concentrations. Rather, their absence as well as the constant low intensity of the tetracarbonyl bands, regardless of the length and intensity of the excitation, is consistent with an oxidation that limits the steady-state amount of W(CO), formed and prevents the formation of W(CO), (n I3) species. The sudden susceptibility to oxidation when the complex achieves the molecularity M(CO),(ads) is attributed to an inability of the photoproduct to complete its coordination shell. As pointed out above, one to three silanol groups are in the immediate vicinity of the photogenerated pentacarbonyl. Consequently, the pentacarbonyl readily completes its coordination shell by binding to a single silanol group and is relatively stable on PVG." The tetracarbonyl, on the other hand, is unable to achieve the same stability. Assuming that there is an average of 5 silanol groups/ 100 AZand that these are present at the corners and center of a 10 X 10 A square, spanning any two of these groups requires M - O (0 being the silanol oxygen) bond lengths of 25 A, considerably larger than known bond lengths.32 Of course, the assumptions that the silanol groups are uniformly distributed is suspect since DRIFT spectra of the glass indicate the presence of both free and associated silanol However, this does not change the fundamental argument that, at some point during the photoinduced decarbonylation, the decarbonylated fragment is unable to complete its coordination shell by binding to the surface silanol groups. At this point, which our data and those of Brenner and c o - w o r k e r ~ indicate ~ ~ + ~ ~ occurs at an average molecularity of M(CO),, the photoproduct becomes significantly more susceptible to oxidation and as a result switches to a photoinduced reaction pathway leading to CH4 evolution. The change in the stoichiometry of CH4 evolution (Figure 5) indicates two reaction pathways. Initially, CHI evolution occurs with a stoichiometric oxidation of the complex, whereas at later times, the evolution of the hydrocarbon becomes independent of metal oxidation. Both reactions are photochemically driven, yet neither involves hydrogenation of coordinated CO. Gas chromatograms show a recovery corresponding to 96 f 2% of the initially coordinated CO, and 13Clabeling establishes that, within an experimental error of 1 1%, mne of the photochemically evolved methane derives from the initially coordinated CO. The evolution of CH, is an impurity effect, but not one that derives from the more conventional impurity sources. Methane evolution from samples prepared by solventless techniques precludes the impregnating solvent, n-hexane, as the carbon source. The absence of higher molecular weight fragments, m/e 1 46, in the mass spectra precludes contaminants that might have been adsorbed during manipulations of the sample on a vacuum line. These would have to be oils or hydrocarbons of sufficiently high molecular weight to withstanding degassing at p Ilo4 Torr and 250 OC. Since the total amount of CH4 evolved falls within the known C impurity level in the glass, 11%,44the carbon source for CH, evolution is attributed to an impurity within the glass matrix. The nature of this impurity is not but its ability to withstand the calcination conditions, in our opinion, precludes any type of hydrocarbon. A carbide would be thermally stable, but in view of the large amount of water initially present in PVG, we would expect that it would react with water and be desorbed as a hydrocarbon during calcination. Since CH, is the only hydrocarbon evolved, we believe that the impurity is a C1 oxide within the glass matrix.43 CH4 derived from an impurity rather than coordinated CO would also account for the dependence of the CH, yield on the pretreatment of the glass. Extensive dehydration of the glass
Simon et al. and calcination at 550 OC under flowing H2 reduce the CH, yield during subsequent photolysis of the impregnated sample. Consider first the stoichiometric conversion of the C1 oxide to CH4. As found in the temperature-programmed decomposition of Cr(C0)6, Mo(CO)6, and W(CO), physisorbed onto silica gel,39940the photoinduced evolution of CH4 is also independent of the initial loadings, which in our experiments correspond to surface coverages ranging from 0.0008 to essentially monolayer coverage. In addition, there is no indication in either the electronic or DRIFT spectra of dimer,
